基于概率神经网络的设备故障诊断及仿真分析

Equipment Fault Diagnosis based on Probabilistic Neural Networks and Simulation Analysis

针对某些难于建立准确数学模型的复杂系统,用神经网络的方法进行故障诊断有其独特的优越性。首先分析了概率神经网络(PNN)的基本结构及其训练算法,建立了某型航空发动机故障分类的概率神经网络模型,通过对该设备故障进行定性诊断,对比分析了概率神经网络与常用的误差反向传播神经网络(BPNN)分类模型对各类故障的分类效果。仿真表明,基于PNN模型的分类方法在分类速度、精度和泛化能力方面均优于基于BPNN的模型,是一种有效的故障分类方法。

Artificial neural network is a useful tool for fault diagnosis of certain complex system that can＇t be mathematically modeled. This paper analyzed the basic theory and algorithm of the probabilistic neural network, and established certain equipment fault classification model based on the PNN and improved BPNN, simulation showed that PNN model outperforms the improved back-propagation neural network model in classification speed, precision and generalization ability. It proves to be an efficient fault classification method.